Automatically Detecting Method for Time-Varying Text Region of Interest

2. The automatically detecting method for time-varying text region of interest according to claim 1, wherein the converted grayscale images in the steps (B03) and (B06) are stored in the historical image database as the grayscale historical images of the screen.

3. The automatically detecting method for time-varying text region of interest according to claim 1, wherein the normalization process in the step (B14) is to binarize values of the pixels of the grayscale image to be in a range of 0 to 1.

4. The automatically detecting method for time-varying text region of interest according to claim 1, wherein the image connection process in the step (B15) is to perform blob analysis on the score map and extract high score blobs.

5. The automatically detecting method for time-varying text region of interest according to claim 1, wherein a filtering process comprises dilation and closing operation based on morphology and is performed to remove low score blocks before the image connection process in the step (B15).

8. The automatically detecting method for time-varying text region of interest according to claim 7, wherein the projection-based search is performed based on a condition that the time-varying text is located between an attribute and an unit, the grayscale image is projected and transformed onto a histogram, and a region, in which consecutive projection values are not zero or lower than a preset value, is searched on the histogram to determine a bounding box region of the attribute and the time-varying text region of interest, and the width of the bounding box region is selected as the width of the time-varying text region of interest, and the height of the bounding box region is equal to the height of the search window.

9. The automatically detecting method for time-varying text region of interest according to claim 7, wherein the Hough transform is configured to perform edge detection on the search window to retrieve bounding box information, transforming the bounding box information to an accumulator map of the Hough transform, and lines are reconstructed by setting the threshold values on the accumulator map, so as to compute the bounding box equal to the input box.

10. The automatically detecting method for time-varying text region of interest according to claim 6, wherein the converted grayscale images in the steps (C09) and (C12) are stored in the historical image database as the grayscale historical images of the screen.

11. The automatically detecting method for time-varying text region of interest according to claim 6, wherein the normalization process in the step (C20) is to binarize values of the pixels of the grayscale image to be in a range of 0 to 1.

12. The automatically detecting method for time-varying text region of interest according to claim 6, wherein the image connection process in the step (C21) is to perform blob analysis on the score map and extract high score blobs.

13. The automatically detecting method for time-varying text region of interest according to claim 6, wherein a filtering process comprises dilation and closing operation based on morphology and is performed to remove low score blocks before the image connection process in the step (C21).